The Anthropic IPO is done. The company behind Claude priced its shares at $83 on Nov. 18, 2026, above the $70 to $78 range it had marketed, and the stock began trading on the Nasdaq Global Select Market under the ticker ANT on Nov. 19. NPR reported the offering raised about $81.3 billion, the largest on record.
Most coverage is about the money. This post is about your website. Claude reads the web, cites pages in its answers and is now used by thousands of businesses, so we read Anthropic's S-1, the registration statement it filed with the SEC on Oct. 26, for what it says about crawling, citations and ads. Then we checked how 85 of the world's biggest sites treat Claude's crawlers today.
This is news, not investment advice. We don't hold a view on the stock.
How did the Anthropic IPO go?
- Price: $83 a share, set on Nov. 18. Anthropic announced the price on X.
- Size: about $81.3 billion raised in the offering, according to NPR and Yahoo Finance. Before pricing, Renaissance Capital reported a planned private placement of $26 billion at the offer price, with buyers including Google, Microsoft and NVIDIA.
- First day: the first trade on Nov. 19 was at $111.50, about 34% above the IPO price, according to Yahoo Finance. The stock faded through the day and closed at $97.17, about 17% above the IPO price, The National reported.
- Structure: Anthropic is a public benefit corporation. Its S-1 says a "financially disinterested" body, the Long-Term Benefit Trust, can elect and remove a majority of the board.
What the S-1 says about how Claude makes money
The filing describes two sources of revenue: usage, which Anthropic calls "pay-for-intelligence" and meters in tokens, and subscriptions to Claude plans. We searched the full text for advertising. Every mention is about Anthropic's own marketing spend or the laws it must follow; the filing describes no advertising business.
That matters to marketers because the two biggest AI assistants are now on different paths. ChatGPT sells ads (we covered how ChatGPT ads are bought). Claude, by its own filing, makes money from usage and plans. If you want your business to show up in Claude's answers, there is no ad slot to buy; you are cited or you aren't.
Claude cites the web, and the S-1 says so
The filing lists Web Search among Claude's developer tools, describing it as giving Claude real-time web information for "grounded, citation-aware responses." Claude.ai, the consumer app, also has web search. Anthropic calls individual users a "top-of-funnel engine" for its business sales: people use Claude at home, then bring it to work.
The business side is big now. Anthropic says more than 11,900 customers paid it $100,000 or more in the 12 months to Sept. 30, 2026, up from more than 950 a year earlier, and that over half of the Fortune 500 are in that group. Revenue for the first half of 2026 was $16.3 billion, against $1.2 billion in the same months of 2025.
For site owners, that means Claude is one more answer engine where buyers ask questions, and getting cited there works the same way as in ChatGPT: be crawlable, answer the question in the first lines, show dates and sources. Our guide to getting cited by ChatGPT applies almost line for line.
What the filing says about crawling your site
The S-1 says Anthropic collects public web data with "web scraping or web crawling techniques" and takes steps to comply with "applicable robots.txt notices" and paywalls or logins, while warning that its efforts "may not always be successful."
It also lists the lawsuits this has brought. Reddit has sued Anthropic over access to Reddit posts. A Quebec news publisher, La Presse, sued in July 2026, and a group of Canadian news publishers filed a claim in October 2026 saying their content was copied from their websites. A US court approved Anthropic's $1.5 billion settlement of the Bartz book copyright class action in July 2026.
Anthropic runs three crawlers, and its help center says each one does a different job:
Anthropic advises blocking by robots.txt, not by IP address, because an IP block also stops its bots from reading your robots.txt. It publishes its crawler IP list for checking traffic.
Our check: 54 of 85 big sites block Claude's training bot
On Nov. 20, 2026 we fetched the robots.txt files of 92 well-known sites (news publishers in the US, UK, Europe, India, Australia and Canada, big recipe, health, finance and tech sites, SEO tools and blog platforms) and read 85 of them; 7 blocked our request. A short Node.js script checked whether each file blocks the whole site, Disallow: /, for each crawler, either by name or through the catch-all User-agent: * group.
Nineteen sites block ClaudeBot but leave Claude-SearchBot open, including Yahoo, Forbes, the Washington Post, the Financial Times, TechCrunch, Business Insider and Medium. That is the setup Anthropic's three-crawler design allows: stay out of training, stay in search answers. The 35 that block Claude-SearchBot include the New York Times, the BBC, CNN, Reuters, The Guardian and Reddit; most name Anthropic's bots one by one.
Big sites also block Anthropic's training bot more often than OpenAI's (54 against 39). Our count covers whole-site blocks only; a site that blocks some folders counts as allowed.
What to do on your own site
- Decide per crawler, not per company. If you want to be cited in Claude but not used for training, block ClaudeBot and allow Claude-SearchBot and Claude-User. If you sell content and want neither, block all three.
- Check your catch-all rule. Several sites in our list block every unnamed bot with
User-agent: *, which also blocks Claude-SearchBot and Claude-User without anyone choosing that.
- Don't block by IP. Anthropic says it can stop the bots from reading your robots.txt at all.
Our free robots.txt generator writes the rules for each AI crawler, and our llms.txt guide covers the other file AI companies discuss.
Why we follow Claude closely
The Search Post is produced with Claude: scheduled Claude routines watch the news feeds, draft posts like this one from primary sources and check the live site every day. We describe the setup in how we run an AI news desk. We have no business relationship with Anthropic.
What we don't know yet
- How much of Claude's search traffic reaches websites as clicks. The filing gives no number for that.
- Whether a public Anthropic will publish usage numbers for Claude's web search each quarter, the way it now must publish revenue.
- How the Reddit and Canadian publisher cases end, and whether a settlement would change how Anthropic's crawlers work.